Experimental Approach to Curve Reconstruction Based on Human Visual perception

Curve reconstruction is the problem of constructing polygonal curve(s) from a set of sample points. Among all the research to solve this problem, visual perception based algorithms, DISCUR and VICUR, come up in recent years as intuitive methods. DISCUR and VICUR connect points into patterns that agree with human visual perception by applying two major Gestalt laws: nearness and smoothness. In this paper, the work in DISCUR and VICUR is extended by conducting an experiment to quantify how these two laws underlie human vision. DOE and ANOVA are used to test the hypotheses about how a connection may be influenced by its neighboring points whereas the multivariable nonlinear regression model is adopted to formulate the influence of Gestalt laws on point connectivity. The experimental results show that the proposed approach is effective.